T. castaneum has a sizable repertoire of immune proteins predicted to participate in various humoral and cellular responses against wounding or infection (Additional data file 1). Like other insects 6719, cuticle and epithelia lining its body surfaces, tracheae and alimentary tract may serve as a physiochemical barrier and local molecular defense by producing antimicrobial peptides and reactive oxygen/nitrogen species (ROS/RNS). While this line of defense may block most pathogens, others enter the hemocoel where a coordinated acute-phase reaction could occur to immobilize and kill the opportunists. This reaction, including phagocytosis, encapsulation, coagulation and melanization, is probably mediated by hemocytes and molecules constitutively present in the circulation. These first responders may not only control minor infections but also call fat body and hematopoietic tissues for secondary responses if necessary. At the molecular level, the following events should take place in all insects, including the beetle: recognition of invading organisms by plasma proteins or cell surface receptors, extra- and intracellular signal transduction and modulation, transcriptional regulation of immunity-related genes, as well as controlled release of defense molecules.

Peptidoglycan recognition proteins (PGRPs) serve as an important surveillance mechanism for microbial infection by binding to Lys- and diaminopimelate-type peptidoglycans of walled bacteria 20. Some Drosophila PGRPs (for example, LC and SA) are responsible for cell-mediated or plasma-based pathogen recognition; others (that is, LB and SB) may hydrolyze peptidoglycans to turn on/off immune responses 2122. In T. castaneum, PGRP-LA, -LC and -LD contain a transmembrane segment; PGRP-SA and -SB are probably secreted; PGRP-LE (without a signal peptide or transmembrane region) may exist in cytoplasm or enter the plasma via a nonclassical secretory pathway. Bootstrap analysis and domain organization clearly indicate that Tribolium and Drosophila PGRP-LEs are orthologs - so far no PGRP-LE has been identified in Anopheles, Bombyx or Apis. Other orthologous relationships (for example, TcPGRP-LC and AmPGRP-LC) are also supported by the phylogenetic analysis (Figure 1). The beetle and mosquito PGRP-LA genes encode two alternative splice forms (PGRP-LAa and -LAb). Like Drosophila and Anopheles, Tribolium PGRP-LA and -LC genes are next to each other in the same cluster. Most of the beetle PGRPs resulted from ancient family diversification that occurred before the emergence of holometabolous insects. In contrast, gene duplication occurred several times in the lineages of mosquito and fly (Figure 1).

Multiple sequence alignment suggests that β-1,3-glucan-recognition proteins (β GRPs) and Gram-negative binding proteins (GNBPs) are descendents of invertebrate β-1,3-glucanases 23. Lacking one or more of the catalytic residues, these homologous molecules do not possess any hydrolytic activity. They are widespread in arthropods and act in part to recognize microbial cell wall components such as β-1,3-glucan, lipoteichoic acid or lipopolysaccharide. We have identified three β GRPs in T. castaneum. Tc-β GRP1 and AgGNBP-B1 through -B5 are closely related and represent a young lineage, whereas Tc-β GRP2 and Tc-β GRP3 belong to an ancient group that arose before the radiation of holometabolous insects (Additional data file 2). Since Drosophila has no β GRP-B and Anopheles has five, the presence of a single gene (encoding Tc-β GRP1) in the beetle can be useful for elucidating function of this orthologous group. In addition to the glucanase-like domain, members of the second group contain an amino-terminal extension of about 100 residues. In Bombyx mori β GRP, this region recognizes β-1,3-glucan also 24. M. sexta β GRP2 binds to insoluble β-1,3-glucan and triggers a serine proteinase cascade for proPO activation 25.

C-type lectins (CTLs) comprise a wide variety of soluble and membrane-bound proteins that associate with carbohydrates in a Ca²⁺-dependent manner 26. Some insect CTLs recognize microorganisms and enhance their clearance by hemocytes 19. Gene duplication and sequence divergence, particularly in the sugar-interacting residues, lead to a broad spectrum of binding specificities for mannose, galactose and other sugar moieties. These proteins associate with microbes and hemocytes to form nodules 27 and stimulate melanization response 28. T. castaneum encodes sixteen CTLs: ten (Tc-CTL1, 2, 4 through 10, and 13) with a single carbohydrate recognition domain and one (Tc-CTL3) with two. Five other proteins, tentatively named Tc-CTL11, 12, 14, 15 and 16, contain a CTL domain, a transmembrane region (except for Tc-CTL11), and other structural modules: CTL11 has three CUB and three EGF; CTL12 has six Ig and three FN3; CTL14 has one LDL_rA, three CUB, ten Sushi, nineteen EGF, two discoidin, one laminin G and one hyalin repeat; CTL15 has one FTP, eleven Sushi and two EFh; CTL16 has one FTP and four Sushi. While lineage-specific expansion of the gene family is remarkable in D. melanogaster and A. gambiae 29, we have not found any evidence for that in T. castaneum (or A. mellifera): Tc-CTL1, 2, 5, 6, 8, 9, 12 through 16 have clear orthologs in the other insect species whereas Tc-CTL7, 10 and 11 are deeply rooted (Additional data file 3).

Galectins are β-galactoside recognition proteins with significant sequence similarity in their carbohydrate-binding sites characteristic of the family. Drosophila DL1 binds to E. coli and Erwinia chrysanthemi 30. Leishmania uses a sandfly galectin as a receptor for specific binding to the insect midgut 31. Tc-galectin1 has two carbohydrate recognition domains; Tc-galectin2 and 3 are orthologous to Am-galectin1 and 2, respectively (Additional data file 4).

All fibrinogen-related proteins (FREPs) contain a carboxy-terminal fibrinogen-like domain associated with different amino-terminal regions. In mammals, three classes of FREPs have been identified: ficolin, tenascins, and microfibril-associated proteins 32. They take part in phagocytosis, wound repair, and cellular adhesion 33. In invertebrates, FREPs are involved in cell-cell interaction, bacterial recognition, and antimicrobial responses 343536. The Tribolium genome contains seven FREP genes, which fall into three groups (Additional data file 5): the expansion of group I yielded four family members: Tc-FREP1 through 4. Sitting next to each other on chromosome 3, these beetle genes encode polypeptides most similar to angiopoietin-like proteins. During angiogenesis, the human plasma proteins interact with tyrosine kinase receptors (for example, Tie) and lead to wound repair and tissue regeneration 37. In group II, Tc-FREP5 is orthologous to Dm-scabrous, which is required for Notch signaling during tissue differentiation 38. Interestingly, Notch is also needed for proper differentiation of Drosophila hemocytes 39. Group III includes Tc-FREP6, Tc-FREP7, Ag-FREP9 and Dm-CG9593. No major expansion has occurred in the beetle or honeybee, in sharp contrast to the situations in the fly and mosquitoes - there are 61 FREP genes in the A. gambiae genome 29.

Thioester-containing proteins (TEPs), initially identified in D. melanogaster 39, contain a sequence motif (GCGEQ) commonly found in members of the complement C3/α 2-macroglobulin superfamily. After cleavage activation, some TEPs use the metastable thioester bond between the cysteine and glutamine residues to covalently attach to pathogens and 'mark' them for clearance by phagocytosis 40. One of the 15 TEPs in Anopheles, Ag-TEP1, plays a key role in the host response against Plasmodium infection and ten other Ag-TEPs are results of extensive gene duplications. This kind of family expansion did not happen in the beetle (or bee): Tribolium encodes four TEPs, perhaps for different physiological purposes. Our phylogenetic analysis supports the following orthologous relationships: TcA-AmA-Ag13-Dm6, TcB-AmB-Ag15-Dm3, and TcC-AmC (Additional data file 6).

Similar to the alternative and lectin pathways for activation of human complements, insect plasma factors play critical roles in pathogen detection, signal relaying/tuning, and execution mechanisms. Serine proteinases (SPs) and their noncatalytic homologs (SPHs) are actively involved in these processes. Some SPs are robust enzymes that hydrolyze dietary proteins; others are delicate and specific - they cleave a single peptide bond in the protein substrates. The latter interact among themselves and with pathogen recognition proteins to mediate local responses against nonself. The specificity of such molecular interactions could be enhanced by SPHs, adaptor proteins that lack proteolytic activity due to substitution of the catalytic triad residues. SPs and SPHs constitute one of the largest protein families in insects 294142. We have identified 103 SP genes and 65 SPH genes in the Tribolium genome, 77 of which encode polypeptides with a SP or SP-like domain and other structural modules. These include thirty SPs and eighteen SPHs containing one or more regulatory clip domains. Clip-domain SPs, and occasionally clip-domain SPHs, act in the final steps of arthropod SP pathways 43. Other recognition/regulation modules (for example, LDL_rA, Sushi, CUB and CTL) also exist in long SPs (>300 residues), some of which act in the beginning steps of SP pathways.

T. castaneum clip-domain proteins are divided into four subfamilies (Figure 2). Even though the catalytic or proteinase-like domains used for comparison were similar in length and sequence, we found subfamily A is composed of SPHs solely whereas subfamilies B, C and D comprise SPs mainly. Apparently, it is easier for SPs to lose activity and become SPHs during evolution than for SPHs to regain catalytic activity. The four groups of SP-related genes may represent lineages derived from ancient evolutionary events since similar subfamilies also exist in Anopheles and Drosophila. Moreover, expansion of individual subfamilies must have occurred several times to account for the gene clusters observed in the Tribolium genome (Figure 2). Evidence for lineage-specific gene duplication and movement is also present in the mosquito and fly genomes 2941. Based on the results of genetic/biochemical analysis performed in other insects 141516194445 and sequence similarity, we are able to predict the physiological functions for some Tribolium clip-domain SPs and SPHs during proPO activation and spätzle processing. For instance, Tc-SPH2, SPH3 or SPH4 (similar to Hd-PPAF2) may serve as a cofactor for Tc-SP7, SP8 or SP10 (putative proPO activating proteinases); Tc-SP44 or SP66 may function like Drosophila persephone 46; Tc-SP136 or SP138 may activate spätzle precursors by limited proteolysis 4445.

Most members of the serpin superfamily are irreversible inhibitors of SPs and, by forming covalent complexes with diffusing proteinases, they ensure a transient, focused defense response 47. There are totally 31 serpin genes in T. castaneum, more than that in D. melanogaster (28), A. gambiae (14) or A. mellifera (7). This number increase is mainly caused by a recent family explosion at a specific genomic location - we have identified a cluster of 16 serpin genes in a small region of 50 kilobases on chromosome 8. These closely related genes constitute a single clade in the phylogenetic tree (Figure 3). Sequence divergence, especially in the reactive site loop region, is anticipated to alleviate the selection pressure imposed by the SP family expansion (Figure 2). Exon duplication and alternative splicing, found in 4 of the 31 serpin genes, also generate sequence diversity and inhibitory selectivity.

Drosophila Toll is a transmembrane protein that binds spätzle and relays developmental and immune signals 48. Resulting from ancient family expansion, a total of five spätzle homologs and eight Toll-like receptors are present in the fly. There are seven Tribolium genes coding for spätzle-like proteins, most of which have putative orthologs in Drosophila and Anopheles (Additional data file 7). Like their ligands, Toll-like proteins have also experienced major family expansion and sequence divergence. The receptors are separated into two clusters, with the fly and beetle Toll-9 located near the tree center (Figure 4). While Toll-6, -7, -8 and -10 from different insect species constitute tight orthologous groups in one cluster, lineage-specific gene duplications have given rise to Drosophila Toll-3 and -4, Anopheles Toll-1 and -5, as well as Tribolium Toll-1 through -4. Located on the same branch with Drosophila Toll, the four Tribolium receptors could play different yet complementary roles in the beetle defense and development. In addition, we have identified eight MD2-related genes in the beetle. Mammalian MD2, Toll-like receptor-4 and CD14 form a complex that recognizes lipopolysaccharides 49. The Anopheles MD2-like receptor regulates the specificity of resistance against Plasmodium berghei 50.

Contrary to the ligand-receptor diversification, components of the intracellular pathway appear to be highly conserved in insects studied so far (Figure 5a). In Drosophila, multimerization of Toll receptors caused by spätzle binding leads to the association of dMyD88, Tube, Pelle, Pellino and dTRAF6 51. With 1:1 orthologs identified in the beetle (as well as the other insects with known genomes), we postulate that a similar protein complex also forms to phosphorylate a cactus-like molecule (Tc02003). The modified substrate protein then dissociates from its partner (Tc07697 or Tc0896), allowing the Rel transcription factors to translocate into the nucleus and activate effector genes (for example, antimicrobial peptides). Functional tests are required to verify the suggested roles of individual components during defense and development in the beetle.

The IMD pathway is critical for fighting certain Gram-negative bacteria in Drosophila. Upon recognition of diaminopimelate-peptidoglycan by PGRPs, the 'danger' signal is transduced into the cell through IMD (Figure 5b). IMD contains a death domain that recruits dFADD (dTAK1 activator) and Dredd (a caspase). Active dTAK1 is a protein kinase that triggers the JNK pathway (through Hep, Basket, Jra and Kay) and Relish phosphorylation (through Ird5 and Kenny). The presence of 1:1 orthologs in T. castaneum strongly suggests that IMD-mediated immunity is conserved in the beetle. Furthermore, the modulation of these pathways may also resemble each other - we have identified putative 1:1 orthologs of IAP2, Tab2 and caspar in the Tribolium genome (Figure 5b).

The transcription of Drosophila TEPs and some other immune molecules is under the control of the JAK-STAT pathway 52. This pathway, triggered by a cytokine-like molecule, Upd3, promotes phagocytosis and participates in an antiviral response. Based on sequence similarity, we predict that the conserved signaling pathway in the beetle is composed of the orthologs of Dm-Domeless, Hopscotch and STAT92 (Figure 5c). However, we have not identified any ortholog of Dm-upd, upd2, or upd3, possibly due to high sequence variation in the cytokine-like proteins.

Phenoloxidases are copper-containing enzymes involved in multiple steps of several immune responses against pathogens and parasites (that is, clot reinforcement, melanin formation, ROS/RNS generation, and microbe killing) 53. Synthesized and released as an inactive zymogen, proPO requires a SP cascade for its cleavage activation. SPHs and serpins ensure that the proteolytic activation occurs locally and transiently in response to infection. We have identified three proPO genes in the Tribolium genome, designated proPO1, 2 and 3. Tc-proPO2 and proPO3 are 98.8% identical in nucleotide sequence and 99.6% identical in amino acid sequence. In the aligned coding regions (2,052 nucleotides long), 21 of the 24 substitutions are synonymous, corresponding to 0.0102 changes/site. These two genes are 530 kb apart and their aligned intron regions are 88.5% identical. Using the relative rate of nucleotide substitutions derived from an analysis of Drosophila alcohol dehydrogenase genes 54, we estimate that Tc-proPO2 and Tc-proPO3 arose by gene duplication approximately 0.6 million years ago. The phylogenetic analysis suggests that such evolutionary events are sporadic for this family: the total numbers of proPO genes in different insect species did not change significantly, except for the malaria mosquito (Additional data file 8). Of the nine Ag-proPO genes, eight arose from gene expansion that occurred early in the mosquito lineage 29, some of which encode phenoloxidases for melanization.

Local production of free radicals is a critical component of the acute-phase oxidative defense, involving nitric oxide synthase, NADPH oxidase, peroxidase, phenoloxidase and other enzymes 5355. Due to the cytotoxicity of ROS and RNS, their conversion and concentrations must be tightly regulated by superoxide dismutases (SODs), glutathione oxidases (GTXs), catalases, thioredoxins, thioredoxin reductases, melanin intermediates, and certain metal ions. Changes in the free radical levels by gene mutation or knock-down affect the fecundity and antimalarial response of the mosquito 56. We have annotated some of these genes in Tribolium, including peroxidases, GTXs, SODs, peroxiredoxins (TPXs) and catalases. T. castaneum GTX1-GTX2 and TPX2-TPX6 gene pairs are results of recent gene duplications, whereas several orthologous relationships have been identified in the SOD and TPX families in the phylogenetic analysis (Additional data file 9).

Coleopteran species have been explored at the biochemical level for various antimicrobial peptides (AMPs) 57. While defensins are present in all insects studied, coleoptericins are related to the attacin/diptericin family of glycine-rich antibacterial peptides in lepidopteran and dipteran species 58. Four defensin genes are detected in the Tribolium genome, three of which are found in a branch containing only coleopteran insects (Figure 6). Tc-defensin4 is in a miscellaneous group containing Odonata, Lepidoptera and Arachnida species. Interestingly, defensins of three other coleopteran insects are in the same branch with the hymenopteran ones. Like the beetle defensins, coleoptericins belong to two phylogenetic groups, with the same separation of species in each group.

With the genome sequence available, we are able to use the other AMP sequences to identify homologous genes that are not specified in beetles. Cecropins were mostly identified in moths and flies - there was only one report on cecropin from a coleopteran species, Acalolepta luxuriosa 11. In Tribolium, we find a single close homolog of the Acalolepta cecropin, although a frame shift in a run of seven adenosines indicate that this is a pseudogene (Tc00499). Closely linked to Tc00499 on chromosome 2 are two genes that encode cecropin-related peptides of unusual structure, with proline- and tyrosine-rich carboxy-terminal extensions (Tc-cecropin2 and Tc00500). These observations indicate that cecropins may widely exist in beetles. Attacins were found only in lepidopteran and dipteran species. We have identified a cluster of three attacin genes (Tc07737-07739) on Tribolium chromosome 4. Although we failed to identify a Drosomycin homolog in the beetle, our search resulted in a low-score hit of a cysteine-rich sequence. The corresponding gene (Tc11324) encodes a 104 residue polypeptide containing 2 whey acidic protein motifs. While mammalian proteins with this motif possess antibacterial activities 59, expression and biochemical analyses are needed to test if the Tribolium protein has a similar function. Due to the presence of species-specific AMPs and severe sequence diversity of these molecules, our homology-based search has probably missed some AMP genes. Should there be a thorough exploration by sequence similarity, biochemical separation and activity assays (not only against Gram-positive and Gram-negative bacteria, but also against yeasts and filamentous fungi), we expect the total number of AMPs (currently 12) in T. castaneum may approach that (20) in D. melanogaster. In addition to these, we have found a cluster of four lysozyme genes in the Tribolium genome (Additional data file 10). Similar but independent family growths have occurred in different insect groups, giving rise to thirteen such genes in Drosophila, eight in Anopheles, three in Apis, and four in Tribolium.

Cellular responses (that is, phagocytosis, nodulation and encapsulation) play key roles in the insect innate immunity 60. In the past few years, breakthroughs have been made in the molecular dissection of these processes 61. Drosophila Peste, Eater, scavenger receptor (SR)-CI, Dscam, TEPs, and PGRP-SC1a seem to be implicated in the phagocytosis. Multiple SR-B genes are present in the Tribolium (16), Drosophila (12) and Anopheles (16) genomes, indicative of important functions of the subfamily. A phylogenetic analysis of the SR-Bs (Figure 7) demonstrates that nearly half of the members arose from ancient gene duplication events - we can easily identify orthologs from different insect species. More recent family expansions in the mosquito 29 and beetle account for the other half of the subfamily. There are two SR-B gene clusters in the Tribolium genome, one of which (TcSR-B14, -B15 and -B16) is located in the same branch containing Dm-peste. In addition to SR-Bs, Drosophila Nimrods are also involved in cellular responses 62. The plasmatocyte-specific NimC1 directly participates in the phagocytosis of bacteria. For Tribolium, all three subclasses are represented: NimA, NimB and NimC, just like in the fly, mosquito and bee. However, unlike the other insects, the syntenic relationship is broken up in the beetle NimC homologs: the two NimC paralogs (Tc02053 and Tc15258) are not closely linked to the NimA and NimB homologs (Tc11427 and Tc11428). In the other insects, the order of nimA, nimB and nimC genes is well conserved.

One characteristic of the innate immune system is that some of its components are transcriptionally up-regulated after a microbial challenge. To acquire evidence that the genes we annotated are involved in defense responses, we have exposed the adult beetles to E. coli, Micrococcus luteus, Candida albicans or Saccharomyces cerevisiae cells and isolated total RNA from the control and treated insects for expression analysis. Real-time PCR experiments indicated that transcript levels of some genes dramatically changed (Figure 8). TcPGRP-SA and TcPGRP-SB mRNA became more abundant after the bacterial infection, whereas the increase was much less significant for TcPGRP-LA, -LE, galectin1 or TEP-C after the C. albicans or M. luteus treatment. Following the Gram-positive bacterial or fungal challenge, we detected some elevations in Tc-cSP66, serpin29 and serpin30 transcripts.

Transcriptional regulation is not limited to pattern recognition molecules or extracellular signal mediators/modulators: we detected differential expression of ligand and their receptors (for example, Tc-spätzle1, Toll-1 through Toll-4, and IMD). mRNA level changes for the latter genes were small except for IMD (Figure 8). Toll-3 and Toll-4 induction after the C. albicans or M. luteus challenge was apparent, although not as notable as IMD. The subtle changes in Toll-1 transcript levels were somewhat different from those of Toll-2, -3 and -4, indicating that there could be functional differences and overlaps in antimicrobial responses for these closely related receptors (Figure 4).

We have also examined genes whose products are plasma proteins directly involved in microbe immobilization or killing. The transcripts of Tc-proPOs, lysozyme1 or lysozyme4 did not significantly change when compared with the controls, whereas those of Tc-lysozyme2 and 3 increased remarkably (Figure 8). The most dramatic increase in mRNA levels occurred in the AMP group of effector molecules, including Tc-attacin2, cecropin3, coleoptericin1, defensin1, and defensin2.

Cluster analysis of the expression patterns has revealed several trends of the transcriptional control of these immune genes. Buffer injected and uninjured adults form one cluster with the lowest mRNA levels, whereas E. coli- and S. cerevisiae-treated insects have the next higher level of overall gene expression (Figure 8). The yeast-injected beetles, instead of grouping with E. coli-treated insects, are found in the same cluster with C. albicans-challenged adults. Interestingly, immune responses toward the opportunistic fungal pathogen are greater than those toward S. cerevisiae, an environmental non-pathogen present in the diet. The responses toward M. luteus and C. albicans were significantly stronger than those towards E. coli, implying that the Toll pathway triggered by the Gram-positive bacteria and filamentous fungi more effectively up-regulated target gene expression than the IMD pathway did, which may be activated by the Gram-negative bacterial infection (Figure 5).

Known defense proteins from other insects were used as queries to perform BLASTP searches of Tcastaneum Glean Predictions (2005.10.11) 63. Protein sequences with E-values lower than 0.1 were listed, and every 5th sequence was retrieved for use as a query for another round of search. Based on the combined lists, respective protein sequences were retrieved, compiled in the order of ascending E-values, and improved by two methods. Firstly, Tcastaneum ESTs (2005.9.20) at the same HGSC site were searched with the corresponding nucleotide sequences to identify possible cDNA clones. The EST sequences were assembled using CAP3 64 and the resulting contigs were used in pairwise comparison 65 to validate the gene predictions. Secondly, retrieved protein sequences were analyzed by CDART 66, PROSITE 67, and SMART 68 to detect conserved domain structures required for specific functions. Necessary changes were made after each step to improve the original predictions. Chromosomal location and exon-intron boundaries for each annotated sequence were acquired from Genboree 69. To locate orthologs not identified by BLASTP, Tribolium Genome Assembly 2.0 70 was searched using TBLASTN. The hits detected were analyzed using multiple gene prediction tools Genescan and Genemark 7172. All curated sequences then were deposited in the annotation database 73 as a part of Tribolium Genome Assembly 2.0.

Unless otherwise specified, full-length Tribolium sequences were aligned with their homologs from other insects, including D. melanogaster, A. gambiae and A. mellifera. The sequences were retrieved from NCBI 74, Flybase 75, or Ensembl 76. Multiple sequence alignments were carried out using ClustalX 77 and Blosum series of weight matrices 78. Phylogenetic trees were constructed based on algorithm of neighbor-joining using PHYLIP 79 or maximum-parsimony using PAUP 80. The divergence time of Tc-proPO2 and proPO3 were calculated using the rate of 1.7 × 10^-8 synonymous substitutions/nucleotide/year derived from the Drosophila species 54.

To study pathogen-induced gene expression, adult red flour beetles (approximately 240 per group) were pricked at the ventral thorax with needles dipped in sterile phosphate-buffered saline or the buffer containing concentrated live E. coli, M. luteus, C. albicans or S. cerevisiae cells. Uninjured and aseptically injured insects were employed as controls. Total RNA samples were extracted from the control and challenged insects (approximately 160 per group) 24 h later, using Micro-to-mid RNA Purification System (Invitrogen, Carlsbad, CA, USA). After DNA removal, each RNA sample (1.0-3.4 μg), oligo(dT) (0.5 μg, 1 μl) and dNTPs (10 mM each, 1 μl) were mixed with diethyl pyrocarbonate-treated H₂O in a final volume of 12 μl, and denatured at 65°C for 5 minutes. First strand cDNA was synthesized for 50 minutes at 42°C using SuperScript Reverse Transcriptase (200 U/μl, 1 μl; Invitrogen) mixed with 5 × buffer (4 μl), 0.1 M dithiothreitol (2 μl), RNase OUT (40 U/μl, 1 μl; Invitrogen) and the denatured RNA sample (12 μl). Specific primer pairs were designed for a total of 35 immunity-related genes (Additional data file 12) using Primer 3 81 with annealing temperatures of 59.5-60.5°C and expected product sizes of 80-150 bp. Each primer pair was located in adjacent exons flanking an intron. Real-time PCR was performed in parallel reactions on 96-well microtiter plates using Taq DNA polymerase (1 U; Roche Applied Sciences, Indianapolis, IN, USA), 1 × buffer, 1 mM dNTP mix, 2 mM MgCl₂, 0.2 μM primers, 1 × SYBR-Green I dye (Applied Biosystems, Foster City, CA, USA) and 10 nM fluorescein. Amplifications were enacted on an iCycler thermal cycler (Bio-Rad, Hercules, CA, USA) with a profile of 95°C for 5 minutes followed by 40 cycles of 94°C for 20 s, 60°C for 30 s, 72°C for 60 s and 78°C for 20 s 82. SYBR green fluorescence was measured during the 78°C step in each cycle and the cycle numbers for each target and control gene were recorded when the fluorescence passed a predetermined threshold. Proper dissociation and correct size of the products were examined by melting curve analysis and agarose gel electrophoresis, respectively. The real-time PCR was repeated twice and, in each of the three experimental replicates, the transcripts were normalized relative to the levels of Tribolium ribosomal protein S3. Averaged transcript abundance values (Ct_control - Ct_target) were then compared across genes and samples using average-linking clustering (Cluster 3.0) and visualized using TreeView 83.